Vm-dev October 2021

vm-dev@lists.squeakfoundation.org

9 participants
59 discussions

[OpenSmalltalk/opensmalltalk-vm] e37c31: Several fixes to Windows makefiles.
by Eliot Miranda 06 Oct '21

06 Oct '21

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: e37c31fba7824cd99b21b4f774ce6b878f777f5d https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/e37c31fba7824cd99b… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2021-10-05 (Tue, 05 Oct 2021) Changed paths: M building/win32x86/common/Makefile M building/win32x86/common/Makefile.msvc M building/win32x86/common/Makefile.msvc.clang.rules M building/win32x86/common/Makefile.msvc.plugin M building/win32x86/common/Makefile.plugin M building/win32x86/common/Makefile.rules M building/win64x64/common/Makefile M building/win64x64/common/Makefile.msvc M building/win64x64/common/Makefile.msvc.clang.rules M building/win64x64/common/Makefile.msvc.plugin M building/win64x64/common/Makefile.msvc.rules M building/win64x64/common/Makefile.plugin M building/win64x64/common/Makefile.rules Log Message: ----------- Several fixes to Windows makefiles. Get the paths right in dep files and don't inlcude phony targets for headers. Don't build an implib for an external plugin. Make the print-foo targets et al PHONY Introduce an always PHONY target to ensure internal and external plugins are always visited by a submake.

1 0

[OpenSmalltalk/opensmalltalk-vm] 344212: Make sure dependency generation on Windows produce...
by Eliot Miranda 05 Oct '21

05 Oct '21

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: 3442124d3c3f1f707129846ea686f720e685695a https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/3442124d3c3f1f7071… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2021-10-05 (Tue, 05 Oct 2021) Changed paths: M building/win32x86/common/Makefile.msvc.clang.rules M building/win32x86/common/Makefile.rules M building/win64x64/common/Makefile.msvc.clang.rules M building/win64x64/common/Makefile.msvc.rules M building/win64x64/common/Makefile.rules Log Message: ----------- Make sure dependency generation on Windows produces dep files relative to $(BUILD) directory.

1 0

[OpenSmalltalk/opensmalltalk-vm] Adds cross-platform compatible virtual-key codes to vm-display-x11 plugin (#599)
by Marcel Taeumel 05 Oct '21

05 Oct '21

Since X11 does not provide a localized version of virtual keys, I mapped the X11 event's `KeySym` to such a "virtual-key code", using `evt.keycode` only as a fall back. It's not perfect but close to what Windows and macOS have. Here is the mapping schema: ``` * 0x00 (unknown virtual key) * 0x01 XK_space (shifted from Latin-1) * 0x02 XK_comma (shifted from Latin-1 OR guessed from xKeycode) * 0x03 XK_period (shifted from Latin-1 OR guessed from xKeycode) * 0x04 .. 0x07 (unused) * * 0x08 .. 0x1f SPECIAL KEYS 1 (from all-mask 0xff00) * 0x20 .. 0x29 NUMERIC ASCII KEYS (shifted from Latin-1) * 0x2a .. 0x43 ALPHABETIC ASCII KEYS (shifted from Latin-1) * * 0x44 US_MINUS (guessed from xKeycode) * 0x45 US_PLUS (guessed from xKeycode) * 0x46 US_LBRACKET (guessed from xKeycode) * 0x47 US_RBRACKET (guessed from xKeycode) * 0x48 US_COLON (guessed from xKeycode) * 0x49 US_QUOTE (guessed from xKeycode) * 0x4a US_TILDE (guessed from xKeycode) * 0x4b US_BACKSLASH (guessed from xKeycode) * 0x4c US_COMMA (guessed from xKeycode) * 0x4d US_PERIOD (guessed from xKeycode) * 0x4e US_SLASH (guessed from xKeycode) * 0x4f US_102 (guessed from xKeycode) * * 0x50 .. 0xff SPECIAL KEYS 2 (from all-mask 0xff00) ``` Here is the new mapping table for Squeak: [new-x11-virtual-keys.1.cs.txt](https://github.com/OpenSmalltalk/opensmallta… See /usr/include/X11/keysymdef.h to understand the "SPECIAL KEYS" (masked 0xff00). The biggest remaining issue is all those "OEM keys", which means the ones you push for "+" or "/" or "~" and so on. *Note that the idea of having a localized version of the virtual key is that you can print it in an application's user interface. Would be really annoying if an app insists on, e.g., Z while you actionally pushed Y.* You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599 -- Commit Summary -- * <a href="https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599/commits/4c64947b…">For keyDown and keyUp events, pass keycodes into the image. Not quite the virtual keys on other platforms because no information about locale.</a> * <a href="https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599/commits/147f7d63…">Prepare skeleton for xkey2sqVirtualKeyCode(). Fix several several naming issues: keyCode -> ascii, codes -> keyNameEntries, nameForKeycode -> nameForKeysym</a> * <a href="https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599/commits/ff3c4265…">Compute a virtual-key code using X11's KeySym and -- as a fall-back -- that event's not-localized KeyCode.</a> -- File Changes -- M platforms/unix/vm-display-X11/sqUnixX11.c (181) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/599

2 7

[OpenSmalltalk/opensmalltalk-vm] 360f5a: Fix a slip in printRegisterState on Apple Silicon ...
by Eliot Miranda 03 Oct '21

03 Oct '21

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: 360f5a5e129713acd2a3859925ccade1555972ad https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/360f5a5e129713acd2… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2021-10-03 (Sun, 03 Oct 2021) Changed paths: M platforms/iOS/vm/Common/Classes/sqSqueakMainApp.m Log Message: ----------- Fix a slip in printRegisterState on Apple Silicon & a warning in crashDumpFile. [ci skip]

1 0

[OpenSmalltalk/opensmalltalk-vm] e9986e: Restore abiulity to compile SoundPlugin on macos v...
by Eliot Miranda 03 Oct '21

03 Oct '21

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: e9986e445149be57287bf4ed4ca8c2579a75a516 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/e9986e445149be5728… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2021-10-03 (Sun, 03 Oct 2021) Changed paths: M platforms/iOS/plugins/SoundPlugin/sqSqueakSoundCoreAudio.m Log Message: ----------- Restore abiulity to compile SoundPlugin on macos versions prior to 11. [ci skip]

1 0

got the git repo cloned, how do I build the dern thing?
by gettimothy 02 Oct '21

02 Oct '21

Hi folks, I haven't touched a vm in a long time. What is the command, on linux to build the linux64 vm? thanks for your time. t

3 5

Re: [Vm-dev] [OpenSmalltalk/opensmalltalk-vm] Fix symbolic C stack backtrace in WIN64 builds. (65d9d81)
by Eliot Miranda 02 Oct '21

02 Oct '21

> On Sep 30, 2021, at 11:21 PM, Marcel Taeumel ***(a)***.***> wrote: > > > N.B. As of cygwin 1.5.25 fopen("crash.dmp","a") DOES NOT WORK! crash.dmp > contains garbled output as if the file pointer gets set to the start of the > file, not the end. So we synthesize our own append mode. > > Hi Eliot. Please update your Cygwin ASAP! It's hardly believeable that the append-mode in fopen is broken in a fairly recent Cygwin installation. > > $ cygcheck -c cygwin > Cygwin Package Information > Package Version Status > cygwin 3.2.0-1 OK > $ cygcheck -f /usr/include/stdio.h > cygwin-devel-3.2.0-1 > This works like a charm and as expected: > > /* File name is hello_world.c */ > #include <stdio.h> > > int main(void) > { > FILE *pFile; > pFile = fopen("crash.dmp","a+t"); > fprintf(pFile, "%s", "Hello!"); > fclose(pFile); > } > PLEASE! Update your Cygwin installation. :-) > I don’t use Cygwin; or rather, I don’t use Cygwin to build. I use gmake, MSVC & clang. Thanks an old comment. I’m not in a position to, nor am I interested in spending the time to, verify if it is still current. > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub, or unsubscribe. > Triage notifications on the go with GitHub Mobile for iOS or Android. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/65d9d814015f68f472…

1 0

[OpenSmalltalk/opensmalltalk-vm] Do not use linux-latest runner, use an older stable version. (#598)
by David T Lewis 01 Oct '21

01 Oct '21

A VM compiled on ubuntu-20.04 (aka latest) will not run on ubuntu-16.04 due to runtime library link issues. Therefore select ubuntu-18.04 as a build runner that is likely to support a wider range of existing Linux systems. Addresses this problem: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.29' not found You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/598 -- Commit Summary -- * <a href="https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/598/commits/783c850d…">Do not use linux-latest runner, use an older stable version.</a> -- File Changes -- M .github/workflows/linux.yml (2) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/598.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/598.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/598

11 41

VM Maker: VMMaker.oscog-eem.3084.mcz
by commits＠source.squeak.org 01 Oct '21

01 Oct '21

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3084.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3084 Author: eem Time: 1 October 2021, 2:45:11.62508 pm UUID: ee7b4320-d292-4786-a3b9-4eadf2628eb7 Ancestors: VMMaker.oscog-mt.3083 Sionara fromIEEE64BitWord:; Kon'nichiwa fromIEEE64Bit: =============== Diff against VMMaker.oscog-mt.3083 =============== Item was changed: ----- Method: CFloatArray>>at: (in category 'accessing') ----- at: offset | address | address := unitSize * offset + self ptrAddress. ^unitSize >= 4 ifTrue: [unitSize = 4 ifTrue: [Float fromIEEE32Bit: (interpreter long32At: address)] + ifFalse: [Float fromIEEE64Bit: (interpreter long64At: address)]] - ifFalse: [Float fromIEEE64BitWord: (interpreter long64At: address)]] ifFalse: [self error: 'unitSize must be 4 or 8']! Item was removed: - ----- Method: Float class>>fromIEEE64BitWord: (in category '*VMMaker-instance creation') ----- - fromIEEE64BitWord: anInteger - | value | - value := self basicNew: 2. - value - basicAt: 1 put: (anInteger bitShift: -32); - basicAt: 2 put: (anInteger bitAnd: 16rFFFFFFFF). - ^value isFinite - ifTrue: [value * 1.0] "reduce to SmallFloat64 if possible" - ifFalse: [value] - - "[| r | - r := Random new. - 100000 timesRepeat: - [| h l f | - h := (r next * 16r100000000) rounded bitAnd: 16rFFFFFFFF. - l := (r next * 16r100000000) rounded bitAnd: 16rFFFFFFFF. - f := Float fromIEEE64BitWord: (h bitShift: 32) + l. - self assert: h = (f basicAt: 1). - self assert: l = (f basicAt: 2)]] timeToRun"! Item was changed: ----- Method: SpurMemoryManager>>floatAtPointer: (in category 'simulation only') ----- floatAtPointer: ptr "See platforms/Cross/vm/sqMemoryAccess.h for the production implementation." <doNotGenerate> + ^Float fromIEEE64Bit: (self long64At: ptr)! - ^Float fromIEEE64BitWord: (self long64At: ptr)!

1 0

VM Maker: Cog-eem.442.mcz
by commits＠source.squeak.org 01 Oct '21

01 Oct '21

Eliot Miranda uploaded a new version of Cog to project VM Maker: http://source.squeak.org/VMMaker/Cog-eem.442.mcz ==================== Summary ==================== Name: Cog-eem.442 Author: eem Time: 1 October 2021, 2:38:40.891132 pm UUID: 91994ee1-d997-4095-9167-2d1b2d6254a0 Ancestors: Cog-eem.441 Sionara fromIEEE64BitWord:; Kon'nichiwa fromIEEE64Bit: =============== Diff against Cog-eem.441 =============== Item was changed: ----- Method: GdbARMAlien>>handleBasicDoubleArithmetic:at: (in category 'floating-point emulation') ----- handleBasicDoubleArithmetic: instr at: pc "Emulate a basic math - add/sub/mul/div - VFP instruction." | rn rd rm vn vm | rn := instr >> 16 bitAnd: 15. rd := instr >> 12 bitAnd: 15. rm := instr bitAnd: 15. + vn := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" + vm := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" - vn := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" - vm := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" "simplest to match the entire instruction pattern rather than mess around shifting and masking and merging" (instr bitAnd: 16rFF00FF0) caseOf: { [16rE200B00 "FMULD"] -> [| r | r := vn * vm. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. [16rE300B00 "FADDD"] -> [| r | r := vn + vm. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. [16rE300B40 "FSUBD"] -> [| r | r := vn - vm. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. [16rE800B00"FDIVD"] -> [| r | r := vn / vm. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord].} otherwise: [self reportPrimitiveFailure]. self pc: pc + 4! Item was changed: ----- Method: GdbARMAlien>>handleExtendedDoubleArithmetic:at: (in category 'floating-point emulation') ----- handleExtendedDoubleArithmetic: instr at: pc "Emulate an extended math - cmp/sqrt/sitod - VFP instruction." | rn rd rm vn vm vd | rn := instr >> 16 bitAnd: 15. rd := instr >> 12 bitAnd: 15. rm := instr bitAnd: 15. + vn := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" + vm := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" - vn := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" - vm := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" "simplest to match the entire instruction pattern rather than mess around shifting and masking and merging" (instr bitAnd: 16rFF00FF0) caseOf: { [16rEB80B80 "FCMPD"] -> ["read rd, compare with rm (ignore rn) and set FPSCR NZCV flags. Sigh" + vd := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rd + 18)). - vd := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rd + 18)). self break]. [16rEB80BC0 "FSITOD"] -> [| r | r := vm asFloat. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. [16rEB10BC0 "FSQRTD"] -> [| r | r := vm sqrt. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. } otherwise: [self reportPrimitiveFailure]. self pc: pc + 4! Item was changed: ----- Method: GdbARMAlien>>handleOneRegTransferDoubleArithmetic:at: (in category 'floating-point emulation') ----- handleOneRegTransferDoubleArithmetic: instr at: pc "Emulate a one-register transfer VFP instruction." | rn rd rm vn vm | rn := instr >> 16 bitAnd: 15. rd := instr >> 12 bitAnd: 15. rm := instr bitAnd: 15. + vn := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" + vm := Float fromIEEE64Bit: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" - vn := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rn + 18)). "Assume accesses fp regs" - vm := Float fromIEEE64BitWord: (self perform: (self registerStateGetters at: rm + 18)). "Assume accesses fp regs" (instr >> 18 bitAnd: 31) caseOf: { [8 "FMULD"] -> [| r | r := vn * vm. self perform: (self registerStateSetters at: rd + 18) with: r asIEEE64BitWord]. [12"FADDD/FSUBD"] -> [self shouldBeImplemented]. [32"FDIVD"] -> [self shouldBeImplemented]. [45"FCMPD"] -> [self shouldBeImplemented]. } otherwise: [self reportPrimitiveFailure]. self pc: pc + 4!

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev October 2021